Transcript A Self-Supervised Terrain Roughness Estimator for Off

A Self-Supervised Terrain Roughness Estimator
for Off-Road Autonomous Driving
David Stavens and Sebastian Thrun
Stanford Artificial Intelligence Lab
Self-Supervised Learning
“Combines” strengths of multiple sensors.
Ultra-Precise, No Range
David Stavens, Sebastian Thrun
Precise, Long Range
Overview




Introduction and Motivation
Classifying Terrain Roughness
Self-Supervised Learning
Experimental Results
David Stavens, Sebastian Thrun
2005 DARPA Grand Challenge
David Stavens, Sebastian Thrun
Velocity Planning for DGC 2005

Mobile robotics traditionally focuses on steering.

But speed is also important.
– Beyond stopping distance and lateral maneuverability.

For Grand Challenge 2005, our vehicle adapted
its speed to terrain conditions, minimizing shock:
–
–
–
–
Increases electrical and mechanical reliability.
Mitigates pose error for laser projection.
Increases traction for improved maneuvers.
Seems to be correlated with slowing on “hard” terrain.
David Stavens, Sebastian Thrun
Velocity Planning for DGC 2005

Simple three state algorithm:
– Drive at speed limit until shock threshold exceeded.
– Slow to bring the vehicle within the shock threshold.
• Uses approx. linear relationship between shock and speed.
• Which is also important for the new work we present.
– Accelerate back to the speed limit.

Discontinuous control problem.
– Hard to solve with conventional control approaches.

We used supervised learning.
David Stavens, Sebastian Thrun
Experiments for DGC 05
David Stavens, Sebastian Thrun
This Talk: Next Logical Step

We expand our online approach to be proactive.
– Our previous approach was entirely reactive.

Difficult to be that precise with laser scanners.
– Hence problems of uncertainty and learning.

Accuracy required for roughness detection
exceeds that required for obstacle avoidance.
– 15cm vs. 2-4cm
David Stavens, Sebastian Thrun
Other Approaches to Velocity Control

Terramechanics: guidance through rough terrain.
– Online assessment only at low speeds.
– High speeds require a priori maps.

Our approach is both online and at high speeds.
– Speeds up to 35 mph.
David Stavens, Sebastian Thrun
CMU’s Preplanning Trailer
David Stavens, Sebastian Thrun
Overview




Introduction and Motivation
Classifying Terrain Roughness
Self-Supervised Learning
Experimental Results
David Stavens, Sebastian Thrun
Acquiring a 3D Point Cloud
David Stavens, Sebastian Thrun
Errors in Pose and Projection
David Stavens, Sebastian Thrun
Z Error vs. Time
David Stavens, Sebastian Thrun
More than t

“Spread” of plot implies more factors than t.

t is also related to:
– Amount/rate of pitching.
– Distance between the two scans.
David Stavens, Sebastian Thrun
Comparing Two Laser Points
pair =
1| z |2 –
3| t |4 –
5| xy distance |6 –
7| dpitch1 |8 – 7| dpitch2 |8 –
9| droll1 |10 – 9| droll2 |10


Seven Features: z, t, xy distance, dpitches, drolls
10 Parameters: 1 2 … 10 (generated with self-supervised learning)
David Stavens, Sebastian Thrun
Combining Multiple Comparisons

n pairs in ascending order.
– Use weighting because resolution of discontinuities is near
resolution of laser. There are not many witness pairs.
n
R =  pair 11i
i=0

This generates a score, R, for that patch of terrain.

But how do we assign target values to R?
David Stavens, Sebastian Thrun
Overview




Introduction and Motivation
Classifying Terrain Roughness
Self-Supervised Learning
Experimental Results
David Stavens, Sebastian Thrun
Self-Supervised Learning
Actual shock when driving over terrain
modifies belief about original laser scan.
Improves classifier for subsequent scans!
David Stavens, Sebastian Thrun
Caveat: Must Correct for Speed
David Stavens, Sebastian Thrun
Mapping from R to Shock
Learn a simple suspension model in
parallel with the classifier:
Rcombined = Rleft
12
+ Rright
12
Rleft and Rright is for the terrain under
each wheel.
David Stavens, Sebastian Thrun
Overview




Introduction and Motivation
Classifying Terrain Roughness
Self-Supervised Learning
Experimental Results
David Stavens, Sebastian Thrun
David Stavens, Sebastian Thrun
David Stavens, Sebastian Thrun
Summary

Road shock provides ground truth for previously
perceived patches of road.

Perception model improves in real-time.

Future terrain assessment is more precise.

A faster route completion time is possible.
– For the same amount of shock.

Works either “offline” or “as you drive.”
– Offline results presented.
David Stavens, Sebastian Thrun
Questions?
David Stavens, Sebastian Thrun